Michael S. Tsirkin wrote:
Quoting Or Gerlitz <[EMAIL PROTECTED]>:
Indeed, today the mad module creates thread per device port, and the cm
module creates thread per cpu, so if the system has 16 cores and two
hcas each with two ports, the IB stack would create 20 threads just for
the sake of the mad and cm modules... I also think it would be good if
the mad module would create one thread and the cm module as well.
Sean - does it make sense to you to change the CM for that matter?
Per-CPU threads like CM does might make sense since they improve data locality.
Sorry but "improve data locality" is not enough information for me to
understand why the IB CM --neeed-- to spawn n kernel threads on my
n-core system, after all its slow path and the data does not moves on
QP1, what's the story here? and if it needs thread-per-cpu, why not use
the system threads/softirqs as does the TCP/IP stack connection mgmt code?
Or.
_______________________________________________
general mailing list
[email protected]
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
